Data Preprocess

Load R Package

First thing first, before starting the project we need to include the package that required for the project.

library(tidyverse) #data preprocess
library(ggplot2) #data visualization
library(caret) #predictive model

Langkah selanjutnya adalah memasukan data kedalam environtment project lalu melihat tipe class data yang ada pada data.

df <- readxl::read_xlsx("dataset/data_sulni.xlsx", sheet=1, col_names = TRUE)
map <- readxl::read_xlsx("dataset/data_sulni.xlsx", sheet=2, col_names = TRUE)

str(df)

## Classes 'tbl_df', 'tbl' and 'data.frame':    131 obs. of  15 variables:
##  $ date                : POSIXct, format: "2008-01-01" "2008-02-01" ...
##  $ p_debt_agriculture  : num  3359 3353 3427 3424 3426 ...
##  $ p_debt_mining       : num  6241 6472 6421 6588 6817 ...
##  $ p_debt_manufacturing: num  18766 18558 19043 19188 19506 ...
##  $ p_debt_electricity  : num  7716 7729 7358 7005 7141 ...
##  $ p_debt_usd          : num  52459 53202 53908 54730 54525 ...
##  $ g_debt_agriculture  : num  1615 1611 1680 1638 1625 ...
##  $ g_debt_mining       : num  701 699 732 717 708 ...
##  $ g_debt_manufacturing: num  2614 2624 2707 2588 2555 ...
##  $ g_debt_electricity  : num  6412 6414 6528 6371 6283 ...
##  $ g_debt_usd          : num  26843 26741 29032 28938 28753 ...
##  $ reserve_position    : num  55999 57125 58987 58770 57464 ...
##  $ us_prime_rates      : num  6 6 5.25 5 5 5 5 5 5 4 ...
##  $ usd_idr             : num  9246 9065 9215 9222 9315 ...
##  $ trend               : num  1 0 1 1 1 0 0 1 1 1 ...

summary(df) #Central of Tendency (Mean, Mode, Median)

##       date                     p_debt_agriculture p_debt_mining  
##  Min.   :2008-01-01 00:00:00   Min.   :3353       Min.   : 6241  
##  1st Qu.:2010-09-16 00:00:00   1st Qu.:4605       1st Qu.:11459  
##  Median :2013-06-01 00:00:00   Median :5901       Median :22882  
##  Mean   :2013-06-01 03:28:51   Mean   :6053       Mean   :19044  
##  3rd Qu.:2016-02-15 12:00:00   3rd Qu.:7398       3rd Qu.:25277  
##  Max.   :2018-11-01 00:00:00   Max.   :9378       Max.   :28898  
##  p_debt_manufacturing p_debt_electricity   p_debt_usd     g_debt_agriculture
##  Min.   :18558        Min.   : 5378      Min.   : 52459   Min.   :1050      
##  1st Qu.:19901        1st Qu.:12517      1st Qu.: 70446   1st Qu.:1309      
##  Median :27370        Median :16966      Median :119102   Median :1558      
##  Mean   :27413        Mean   :16304      Mean   :112362   Mean   :1546      
##  3rd Qu.:34059        3rd Qu.:19657      3rd Qu.:149778   3rd Qu.:1778      
##  Max.   :37268        Max.   :29843      Max.   :169622   Max.   :2197      
##  g_debt_mining   g_debt_manufacturing g_debt_electricity   g_debt_usd   
##  Min.   :399.5   Min.   : 415.0       Min.   :3295       Min.   :26741  
##  1st Qu.:455.8   1st Qu.: 769.6       1st Qu.:3567       1st Qu.:37060  
##  Median :632.4   Median :1290.3       Median :4476       Median :49782  
##  Mean   :630.8   Mean   :1476.7       Mean   :4757       Mean   :53042  
##  3rd Qu.:788.9   3rd Qu.:2209.1       3rd Qu.:5762       3rd Qu.:69094  
##  Max.   :869.4   Max.   :2706.5       Max.   :6838       Max.   :86250  
##  reserve_position us_prime_rates     usd_idr          trend       
##  Min.   : 50182   Min.   :3.250   Min.   : 8500   Min.   :0.0000  
##  1st Qu.: 89175   1st Qu.:3.250   1st Qu.: 9281   1st Qu.:0.0000  
##  Median :107543   Median :3.250   Median :11360   Median :1.0000  
##  Mean   : 99011   Mean   :3.832   Mean   :11264   Mean   :0.5649  
##  3rd Qu.:114376   3rd Qu.:4.375   3rd Qu.:13325   3rd Qu.:1.0000  
##  Max.   :131980   Max.   :6.000   Max.   :15202   Max.   :1.0000

str(map)

## Classes 'tbl_df', 'tbl' and 'data.frame':    5124 obs. of  9 variables:
##  $ category_1: chr  "I. Barang dagangan umum" "I. Barang dagangan umum" "I. Barang dagangan umum" "I. Barang dagangan umum" ...
##  $ category_2: chr  "A. Hasil pertanian" "A. Hasil pertanian" "A. Hasil pertanian" "A. Hasil pertanian" ...
##  $ category_3: chr  "Biji coklat" "Udang" "Biji kopi" "Ikan dan lain-lain" ...
##  $ month     : num  1 1 1 1 1 1 1 1 1 1 ...
##  $ year      : num  2013 2013 2013 2013 2013 ...
##  $ export_usd: chr  "32636.12" "84387.33" "78193.47" "78708.149999999994" ...
##  $ import_usd: chr  "6323.33" "2102.91" "1444.29" "13489.63" ...
##  $ date      : chr  "1/1/2013" "1/1/2013" "1/1/2013" "1/1/2013" ...
##  $ count     : num  1 2 3 4 5 6 7 8 9 10 ...

summary(map)

##   category_1         category_2         category_3            month      
##  Length:5124        Length:5124        Length:5124        Min.   : 1.00  
##  Class :character   Class :character   Class :character   1st Qu.: 3.75  
##  Mode  :character   Mode  :character   Mode  :character   Median : 6.50  
##                                                           Mean   : 6.50  
##                                                           3rd Qu.: 9.25  
##                                                           Max.   :12.00  
##       year       export_usd         import_usd            date          
##  Min.   :2013   Length:5124        Length:5124        Length:5124       
##  1st Qu.:2014   Class :character   Class :character   Class :character  
##  Median :2016   Mode  :character   Mode  :character   Mode  :character  
##  Mean   :2016                                                           
##  3rd Qu.:2018                                                           
##  Max.   :2019                                                           
##      count     
##  Min.   :   1  
##  1st Qu.:1282  
##  Median :2562  
##  Mean   :2562  
##  3rd Qu.:3843  
##  Max.   :5124

Data Cleaning

Menetapkan class yang sesuai dengan data.

map$date <- dmy(map$date)
map$export_usd <- as.numeric(map$export_usd)
map$import_usd <- as.numeric(map$import_usd)

summary(map)

##   category_1         category_2         category_3            month      
##  Length:5124        Length:5124        Length:5124        Min.   : 1.00  
##  Class :character   Class :character   Class :character   1st Qu.: 3.75  
##  Mode  :character   Mode  :character   Mode  :character   Median : 6.50  
##                                                           Mean   : 6.50  
##                                                           3rd Qu.: 9.25  
##                                                           Max.   :12.00  
##                                                                          
##       year        export_usd        import_usd           date           
##  Min.   :2013   Min.   :      0   Min.   :      0   Min.   :2013-01-01  
##  1st Qu.:2014   1st Qu.:  15852   1st Qu.:   3414   1st Qu.:2014-09-23  
##  Median :2016   Median :  53945   Median :  35964   Median :2016-06-16  
##  Mean   :2016   Mean   : 206409   Mean   : 191644   Mean   :2016-06-16  
##  3rd Qu.:2018   3rd Qu.: 220041   3rd Qu.: 141000   3rd Qu.:2018-03-08  
##  Max.   :2019   Max.   :2268562   Max.   :3253121   Max.   :2019-12-01  
##                 NA's   :127       NA's   :148                           
##      count     
##  Min.   :   1  
##  1st Qu.:1282  
##  Median :2562  
##  Mean   :2562  
##  3rd Qu.:3843  
##  Max.   :5124  
##

Mencari tahu apakah ada kekosongan data.

sum(is.na(map$export_usd))

## [1] 127

sum(is.na(map$import_usd))

## [1] 148

map[is.na(map)] <- 0
summary(map)

##   category_1         category_2         category_3            month      
##  Length:5124        Length:5124        Length:5124        Min.   : 1.00  
##  Class :character   Class :character   Class :character   1st Qu.: 3.75  
##  Mode  :character   Mode  :character   Mode  :character   Median : 6.50  
##                                                           Mean   : 6.50  
##                                                           3rd Qu.: 9.25  
##                                                           Max.   :12.00  
##       year        export_usd        import_usd           date           
##  Min.   :2013   Min.   :      0   Min.   :      0   Min.   :2013-01-01  
##  1st Qu.:2014   1st Qu.:  13842   1st Qu.:   2805   1st Qu.:2014-09-23  
##  Median :2016   Median :  51654   Median :  34283   Median :2016-06-16  
##  Mean   :2016   Mean   : 201293   Mean   : 186109   Mean   :2016-06-16  
##  3rd Qu.:2018   3rd Qu.: 212219   3rd Qu.: 135138   3rd Qu.:2018-03-08  
##  Max.   :2019   Max.   :2268562   Max.   :3253121   Max.   :2019-12-01  
##      count     
##  Min.   :   1  
##  1st Qu.:1282  
##  Median :2562  
##  Mean   :2562  
##  3rd Qu.:3843  
##  Max.   :5124

Data Visualization

Basic Plotting

p1 <- ggcorr(df, label = T,label_size = 2.9)

df$us_prime_rates <- as.factor(df$us_prime_rates)

p2 <- ggplot(df, aes(x=date, y=usd_idr, color=us_prime_rates)) + 
  geom_jitter() 

p3 <- ggplot() + 
  geom_line(data=df, aes(x=date, y=usd_idr), color = "red") +
  geom_line(data=df, aes(x=date, y=g_debt_electricity), color = "blue") + 
  geom_line(data=df, aes(x=date, y=p_debt_electricity), color = "green") + 
  xlab('Dates') + 
  ylab('value') + labs(colour = "debt type")

p1

p2

p3

Interactive Plotting

ggplotly(p3)

map1 <- map %>%
  filter(category_2 == "A. Hasil pertanian")
map2 <- map %>%
  filter(category_2 == "B. Hasil industri")
map3 <- map %>%
  filter(category_2 == "C. Hasil pertambangan")

p_1 <- ggplot(
  map, 
  aes(x = export_usd/1000, y=import_usd/1000, colour = factor(category_3))
  ) +
  geom_point(show.legend = FALSE, alpha = 0.7) +
  scale_color_viridis_d() +
  scale_size(range = c(2, 12)) +
  labs(x = "export_usd", y = "import_usd", colour="category_3") +
  facet_wrap(~category_2)

animation1 <- p_1  +
  transition_time(date) +
  labs(title = "Year: {frame_time}") 


p_2 <- ggplot(
  map1, 
  aes(x = export_usd/1000, y=import_usd/1000, colour = factor(category_3))
  ) +
  geom_point(show.legend = TRUE, alpha = 0.7) +
  scale_color_viridis_d() +
  scale_size(range = c(2, 12)) +
  labs(x = "export_usd", y = "import_usd", colour=" ") + 
  theme(legend.position="bottom")

animation2 <- p_2  +
  transition_time(date) +
  labs(title = "Year: {frame_time}") 

animate(animation1, fps=5)

animate(animation2, fps=5)

Predictive Model (Projection)

Logistic Regression

set.seed(50) 
train.control <- trainControl(method = "cv", number = 10)
df$trend <- as.factor(df$trend)
# Train the model
model <- caret::train(trend ~p_debt_mining + p_debt_manufacturing + g_debt_mining + 
    g_debt_manufacturing + g_debt_electricity + g_debt_usd + 
    reserve_position + us_prime_rates, data = df, method = "vglmAdjCat",
               trControl = train.control)

# Summarize the results
print(model)

## Adjacent Categories Probability Model for Ordinal Data 
## 
## 131 samples
##   8 predictor
##   2 classes: '0', '1' 
## 
## No pre-processing
## Resampling: Cross-Validated (10 fold) 
## Summary of sample sizes: 117, 118, 118, 117, 117, 119, ... 
## Resampling results across tuning parameters:
## 
##   parallel  Accuracy   Kappa    
##   FALSE     0.6418396  0.2739485
##    TRUE     0.6418396  0.2739485
## 
## Tuning parameter 'link' was held constant at a value of loge
## Accuracy was used to select the optimal model using the largest value.
## The final values used for the model were parallel = FALSE and link = loge.

SULNI Data Visualization

Group_9

2020-03-06